scsh-users
[Top] [All Lists]

Re: scsh server dying with broken socket

To: Ed Kademan <kademan@phz.com>
Subject: Re: scsh server dying with broken socket
From: Martin Gasbichler <gasbichl@informatik.uni-tuebingen.de>
Date: Fri, 03 Jan 2003 21:36:01 +0100
Cc: scsh-news@zurich.ai.mit.edu
>>>>> "Ed" == Ed Kademan <kademan@phz.com> writes:

Ed> Can someone help me with a client-server application I have written?
Ed> I have attached some code below with a description of the problem in
Ed> the opening comments.  Maybe you can tell what I am doing wrong just
Ed> from looking at those comments.

Ed> ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
Ed> ;;; $Id: test-server.scm,v 1.6 2003/01/03 15:54:08 kademan Exp $
Ed> ;;; 
Ed> ;;; This program is a simplified version of a client-server system I
Ed> ;;; have written.  It demonstrates that if the client process dies
Ed> ;;; between the time the server recognizes the connection and the time
Ed> ;;; it tries to write the information out to the associated port the
Ed> ;;; server dies as well, with a "Broken pipe" error.  The error
Ed> ;;; appears to be untrappable.  I want to trap the error---or prevent
Ed> ;;; it in the first place.

You cannot catch this exception with with-errno-handler but only with
the general with-handler from the handle structure because the
exception comes from the S48 I/O system not from scsh.

This is clearly awkward and needs to be fixed.

-- 
Martin

<Prev in Thread] Current Thread [Next in Thread>